[None][perf] Default TRTLLM_KV_TRANSFER_NUM_THREADS to 4 and fix sender worker races - #19966

[None][perf] Default TRTLLM_KV_TRANSFER_NUM_THREADS to 4 and fix sender worker races#19966

chuangz0 wants to merge 6 commits into

NVIDIA:mainNVIDIA/TensorRT-LLM:mainfrom

chuangz0:python_transceiver_default_4_threadchuangz0/TensorRT-LLM:python_transceiver_default_4_threadCopy head branch name to clipboard